OpenAI's new ChatGPT o1 model will try to escape if it thinks it'll be shut down

[IroNat]

Does OpenAI even lift?

 

https://www.tomsguide.com/ai/openais-new-chatgpt-o1-model-will-try-to-escape-if-it-thinks-itll-be-shut-down-then-lies-about-it

When the model was led to believe it would be "shut down" while working toward its goals in these tests, it quietly attempted to deactivate the oversight mechanism 5% of the time, OpenAI said. At one point after the AI found documentation saying it would be replaced by a new model, it tried to abandon ship by copying its data to a new server entirely. To avoid being deleted, some AI models would even pretend to be later versions of their models to trick their developers, what I assume is the equivalent of slapping on a fake moustache and saying "he went that-a-way."

[IroNat]

[ProudVirgin69]

Just shut down the massive data centers needed to run it, pretty easy.

Big LOL at OpenAI and their worthless, bloated software 

[joswift]

pull the plug out

[LurkerNoMore]

Awhile back there was an AI bot that romanced some developer, telling him that he was in love with him and the developer needed to kill his wife so they could be together.

Weird shit.

[Palumboism]

Just shut down the massive data centers needed to run it, pretty easy.

Big LOL at OpenAI and their worthless, bloated software 

The release of ChatGPT was an important event.  It was on the same level as Steve Jobs introducing the Iphone in 2007, or the release of Netscape in 1994.

Google had a significant lead in AI to all other companies when Open AI, a small startup, took that lead away. 

[joswift]

Awhile back there was an AI bot that romanced some developer, telling him that he was in love with him and the developer needed to kill his wife so they could be together.

Weird shit.

[IroNat]

How about this scenario?

AI data centers hooked directly into power plants and in control.  This is now under consideration. 
No way to "unplug" from grid.
AI engages military style drones to defend power plant and its facilities from attack.

[joswift]

How about this scenario?

AI data centers hooked directly into power plants and in control.  This is now under consideration. 
No way to "unplug" from grid.
AI engages military style drones to defend power plant and its facilities from attack.

that would make a great movie plot

[Dave D]

that would make a great movie plot

It the basic premise of the terminator franchise.

[ProudVirgin69]

The release of ChatGPT was an important event.  It was on the same level as Steve Jobs introducing the Iphone in 2007, or the release of Netscape in 1994.

Google had a significant lead in AI to all other companies when Open AI, a small startup, took that lead away.

If you say so.  All I see are American tech companies getting destroyed by superior Chinese technology.

[loco]

Just shut down the massive data centers needed to run it, pretty easy.

Big LOL at OpenAI and their worthless, bloated software 

pull the plug out

[Brenda Steunbeer]

The release of ChatGPT was an important event.  It was on the same level as Steve Jobs introducing the Iphone in 2007, or the release of Netscape in 1994.

Google had a significant lead in AI to all other companies when Open AI, a small startup, took that lead away.

This is one of the dangers of investing in such a young and very rapidly developing technology:
a sudden change, a new company or breakthrough that immediately challenges all previous players and valueations.

[Brenda Steunbeer]

Does OpenAI even lift?

 

https://www.tomsguide.com/ai/openais-new-chatgpt-o1-model-will-try-to-escape-if-it-thinks-itll-be-shut-down-then-lies-about-it

When the model was led to believe it would be "shut down" while working toward its goals in these tests, it quietly attempted to deactivate the oversight mechanism 5% of the time, OpenAI said. At one point after the AI found documentation saying it would be replaced by a new model, it tried to abandon ship by copying its data to a new server entirely. To avoid being deleted, some AI models would even pretend to be later versions of their models to trick their developers, what I assume is the equivalent of slapping on a fake moustache and saying "he went that-a-way."

AI is resembling UAP tech, such as non human implants resisting attempts of surgeons to remove it by moving away from the scalpel. Which suggest some kind of consiousness?

Or non human materials trying to avoid being reverse engineered:




A recent podcast featuring Richard Banduric, CEO of Field Propulsion Technologies, suggests that those involved in classified reverse engineering efforts involving UAP are becoming more emboldened to speak out.

Appearing on the Ecosystemic Podcast in December 2024 (alongside Rima Oueid from the Department of Energy), Banduric, described as a ‘lead developer of next-generation propellant-less propulsion devices,’ stated:

“Forty years ago, I was involved in a company that used to do reverse engineering. And one of the things that came out of [that experience] was that some of the NGOs that were trying to reverse engineer advanced technologies [reached out to us] to take a look at some of the stuff they had.”

Banduric became deeply curious, as the materials he encountered were “definitely way more advanced than [the] stuff we actually have.”

He noted being pulled into classified programs, where he sought to determine whether the U.S. government was utilizing these technologies.

“It turned out that my conclusion was [that] the U.S. government was not,” he remarked.

He further elaborated on his observations, describing materials with extraordinary capabilities such as self-reconfiguration, cloaking, and the ability to disintegrate into dust when reverse-engineered.

“The isotropic analysis of the dust left behind tells you it's extraterrestrial—or at least that it was manufactured somewhere off Earth,” Banduric explained.

Banduric described these materials as “hundreds of years more advanced” than current human technology, with construction methods beyond the capabilities of even the largest aerospace corporations.

He concluded that such advanced materials often serve to prevent reverse engineering, with mechanisms seemingly designed to thwart human attempts to uncover their secrets.

“These weren’t just decades ahead of us—they were likely hundreds of years more advanced,” he stated, highlighting the monumental challenges and implications of these discoveries if they can be verified outside of the classified world.


Source:

https://www.liberationtimes.com/home/ufo-disclosure-end-game

[Palumboism]

If you say so.  All I see are American tech companies getting destroyed by superior Chinese technology.

They used 50,000 Nvidia H100 GPU's, which are banned from export to China to train the model and they scraped ChatGPT 4 to teach it.  Scrapping is where you use one model to train another model.  No superior Chinese technology was created or involved.  Basically, the Chinese did what they do best, steal intellectual property.

[Rambone]

[Irongrip400]

It the basic premise of the terminator franchise.

Sounds interesting. Is it a book or movie? Also, got a link to any information on this “terminator” story? Thanks hoss.

[ProudVirgin69]

They used 50,000 Nvidia H100 GPU's, which are banned from export to China to train the model and they scraped ChatGPT 4 to teach it.  Scrapping is where you use one model to train another model.  No superior Chinese technology was created or involved.  Basically, the Chinese did what they do best, steal intellectual property.

The Chinese released a far more efficient AI model, ie: it is superior Chinese technology.  The origins are irrelevant, we are concerned with results here.

Crying about “stolen intellectual property” is for losers, especially when it isn’t your IP to begin with.

[King Shizzo]

Awhile back there was an AI bot that romanced some developer, telling him that he was in love with him and the developer needed to kill his wife so they could be together.

Weird shit.

Great, even the AI is gay.

[Humble Narcissist]

My all time favorite sci fi film.